Live Prediction of Fraudulent Claims in Insurance Company

Live @ http://mandbinsurance-env.eba-qbzfhkdt.us-east-1.elasticbeanstalk.com/

(Deployed Model in a private repo (want to view :- Contact me on folasadetosin0203@gmail.com))

Overview

This project focuses on utilizing machine learning techniques to predict fraudulent claims in the insurance industry. The aim is to develop a predictive model that can accurately identify potential fraudulent activities, thereby enabling the company to take necessary actions to mitigate risks and losses.

Project Scope

The project encompasses the following key aspects:

Data Collection:

Gathering relevant data related to insurance claims, including historical claim records, customer information, policy details, etc.

Data Preprocessing:

Cleaning and preprocessing the collected data to ensure consistency, accuracy, and compatibility with machine learning algorithms. This involves handling missing values, encoding categorical variables, and standardizing numerical features.

Exploratory Data Analysis (EDA):

Conducting comprehensive EDA to gain insights into the data distribution, correlations, anomalies, and patterns. This step is crucial for identifying potential features and understanding the underlying dynamics of fraudulent claims.

Feature Engineering:

Creating new features or transforming existing ones to enhance the predictive power of the model. Feature engineering plays a vital role in extracting meaningful information from raw data and improving the model's performance.

Model Development:

Building machine learning models to predict the likelihood of a claim being fraudulent. Various algorithms such as logistic regression, decision trees, random forests, and gradient boosting will be explored and evaluated to identify the most suitable model for the task.

Model Evaluation:

Assessing the performance of the developed models using appropriate evaluation metrics such as accuracy, precision, recall, F1-score, and ROC-AUC. This step involves fine-tuning the models and selecting the best-performing one for deployment.

Live Prediction:

Implementing the final model into a live prediction system where new insurance claims can be processed in real-time. This system will provide instant feedback on the likelihood of a claim being fraudulent, enabling timely intervention and decision-making by the company.

Technologies Used

Python Jupyter Notebook Pandas NumPy Scikit-learn Matplotlib Seaborn

Contributing

Contributions to this project are welcome! If you have any suggestions, bug fixes, or enhancements, feel free to get in touch on folasadetosin0203@gmail.com